Magnetic resonance fingerprinting: : accuracy, repeatability and implementation in clinical practice
Magnetic Resonance Fingerprinting (MRF) was recently introduced as a novel approach to fast quantitative Magnetic Resonance Imaging (MRI) that allows simultaneous and efficient measurement of multiple tissue properties during one single, rapid acquisition. The aim of this work was to estimate the stability and accuracy of the MRF framework in an imaging phantom and non-pathologic brain tissue from healthy volunteers and a large patient collective by measurements of longitudinal (T1) and transverse (T2)relaxation times at field strengths of 1.5 and 3 T. A goal was also to investigate the options to integrate MRF into routine workflows of neuroimaging in the clinical setting.
Repetitive phantom scans yielded stable and reproducible results across a wide range of relaxometry values with coefficients of variation less than 7 % for T1 and less than 14 % for T2 estimates at both field strengths. Linear correlation of MRF results and relaxometry data obtained by Nuclear Magnetic Resonance (NMR) measurements validated the accuracy of the MRF method in vitro. MRF relaxometry measurements of various brain tissues in a study population of 10 healthy volunteers were in sufficient agreement with conventional quantitative imaging techniques and previous MRF data found in literature. The inter-individual variability of relaxometry measurements obtained from multiple structures of white matter, basal nuclei and thalamus showed maximum values of 6 % for T1 and 15 % for corresponding T2 measurements at 1.5 and 3 T, confirming an adequate repeatability of the MRF method in vivo. After validation of phantom and volunteer measurements, MRF scans were successfully integrated into the clinical workflow. Within a time period of six month, 92 patients with either previously diagnosed neurological pathologies or suspected diagnosis of brain pathologies that were scheduled for a routine clinical MRI protocol at 3 T received an additional MRF brain scan. From anatomical regions that were not infiltrated by pathology MRF relaxometry values were compared across subjects and to corresponding results in healthy volunteers. In general, T1 and T2 estimates obtained from patient data showed higher variation coefficients than those in healthy subjects. This may be primarily attributed to the wider spread in age and larger impact of motion artifacts in the patient cohort. The comparison of relaxometry values from a sub-group of 10 patients at the same age range as the healthy volunteers revealed similar low coefficients of variation as found in the volunteer cohort, thus confirming the long-term stability of the MRF measurements over several months.
Obtained relaxometry data from multiple healthy brain structures may contribute to a normative database of MRF relaxometry estimates that could help to establish quantitative imaging biomarkers for improvement of diagnostics in neuroimaging.